Advanced Classification of Diabetic Foot Ulcers Using Custom and Deep Learning Models

Diabetic Foot Ulcers (DFUs) pose a significant health threat, often leading to severe complications if not promptly diagnosed and treated. This research introduces a novel approach for DFU classification by developing and evaluating both custom and established deep learning models. Our custom CNN model is meticulously designed to balance efficiency and performance, particularly in resource-constrained environments, and is optimized for extracting intricate patterns and features. We applied various optimizers, including SGD, RMSprop, Adam, and Nadam, all with a learning rate of 0.0045, with the Adam optimizer achieving an exceptional accuracy of $\mathbf{9 6. 2 1 \%}$. Comprehensive evaluations on a standardized dataset of 800 images demonstrate the superior performance of our custom model compared to VGG16, MobileNet, ResNet, and DenseNet models. This study highlights the potential of advanced deep learning techniques to significantly enhance DFU classification, ultimately contributing to better patient outcomes and clinical practices.